Gray matter structural networks are associated with cardiovascular risk factors in healthy older adults

Abstract

While recent 'big data' analyses discovered structural brain networks that alter with age and relate to cognitive decline, identifying modifiable factors that prevent these changes remains a major challenge. We therefore aimed to determine the effects of common cardiovascular risk factors on vulnerable gray matter (GM) networks in a large and well-characterized population-based cohort. In 616 healthy elderly (258 women, 60-80 years) of the LIFE-Adult-Study, we assessed the effects of obesity, smoking, blood pressure, markers of glucose and lipid metabolism as well as physical activity on major GM-networks derived using linked independent component analysis. Age, sex, hypertension, diabetes, white matter hyperintensities, education and depression were considered as confounders. Results showed that smoking, higher blood pressure, and higher glycated hemoglobin (HbA1c) were independently associated with lower GM volume and thickness in GM-networks that covered most areas of the neocortex. Higher waist-to-hip ratio was independently associated with lower GM volume in a network of multimodal regions that correlated negatively with age and memory performance. In this large cross-sectional study, we found selective negative associations of smoking, higher blood pressure, higher glucose, and visceral obesity with structural covariance networks, suggesting that reducing these factors could help to delay late-life trajectories of GM aging.

Keywords: Alzheimer's disease; Independent component analysis; aging; brain structure; gray matter modifiers; structural covariance.

Figures

Figure 1.

Flow chart of the study. Out of 985 older adults free of stroke, 369 were excluded due to medication intake, brain pathology, missing covariates, non-usable MRI scans, or non-intact cognition, leaving 616 participants for main analyses. Out of this sample, 516 participants had physical activity information.

Figure 2.

Bivariate correlations among independent components (IC), cardiovascular risk factors, confounders, and verbal memory score. Significant associations (Spearman's correlations,p < 0.05) are color-coded in red-shaded (positive) and blue (negative). CV: cardiovascular; WMH: white matter hyperintensities; TIV: total intracranial volume; APOE-e4: apolipoprotein E epsilon-4 carrier status; BMI: body mass index; WHR: waist-to-hip ratio; HbA1c: glycated hemoglobin; HDL: high-density lipoprotein.

Figure 3.

In two global networks, lower gray matter thickness (IC2, a) and volume (IC7, b) were associated with smoking (a) and higher blood pressure (b). Scatter plots show the individual’s loading (black dots) and the group’s median with 95% SE or linear fit. Colors indicate positive (red/yellow) or negative (blue/light-blue) co-variations within the network (z > 4), maps are drawn on a standard brain.

Figure 4.

Higher fasting serum levels of HbA1c were associated with lower cortical thickness of IC2 (a) and IC5 (b). Scatter plots show the individual’s loading on each network (black dots) and linear fit. Colors indicate positive (red/yellow) or negative (blue/light-blue) co-variations within the network (z > 4), maps are drawn on a standard brain.

Figure 5.

Higher waist-to-hip ratio was associated with lower gray matter volume in a network of multimodal regions (IC3, a, c) that also correlated negatively with age and memory performance (b). Scatter plot shows the individual’s loading on the network and linear fit. Colors indicate positive (red/yellow) or negative (blue/light-blue) co-variations within the network (z > 11), maps are drawn on a standard brain.